All of lore.kernel.org
 help / color / mirror / Atom feed
From: srinivas pandruvada <srinivas.pandruvada@linux.intel.com>
To: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Aaron Rainbolt <arainbolt@kfocus.org>,
	kernel-team@lists.ubuntu.com,  lenb@kernel.org,
	linux-acpi@vger.kernel.org, linux-pm@vger.kernel.org,
	 rjw@rjwysocki.net, mmikowski@kfocus.org
Subject: Re: [BUG] intel_pstate: CPU frequencies miscalculated/incorrectly detected on Arrow Lake hardware
Date: Fri, 05 Dec 2025 11:50:07 -0800	[thread overview]
Message-ID: <c7bdb6cb02437d1ab41495d526c3d2ee3f1f7c60.camel@linux.intel.com> (raw)
In-Reply-To: <CAJZ5v0hUc3oz=Z8UUv3n4rY3on1vZpCtSdKjgOFp+OWYC5D9tw@mail.gmail.com>

On Fri, 2025-12-05 at 20:28 +0100, Rafael J. Wysocki wrote:
> On Fri, Dec 5, 2025 at 8:11 PM srinivas pandruvada
> <srinivas.pandruvada@linux.intel.com> wrote:
> > 
> > Hi Aaron,
> > 
> > On Wed, 2025-12-03 at 11:38 -0600, Aaron Rainbolt wrote:
> > > On Wed, 26 Nov 2025 17:00:31 -0600
> > > Aaron Rainbolt <arainbolt@kfocus.org> wrote:
> > > 
> > > 
> > > 
> > ...
> > 
> > 
> > 
> > There are too many platforms here, don't have time to get to all.
> > So
> > let's focus on one first.
> > 
> > ## Clevo | X580WNT-G  | Ultra 9 275HX | 6.18.0-061800rc7-generic
> > 
> > Summary:
> > There is no scaling factor related issue as reported before on
> > ASROCK
> > platform before. The achieved maximum frequencies are correct.
> > 
> > The base_freq display is wrong (see below because of BIOS config).
> > The cpuinfo_max_freq wrong display is not related to scaling factor
> > but
> > something else
> > 
> > 
> > To check, I need dump of:
> > m=$(getconf _NPROCESSORS_ONLN); for ((i=0; i<m; i++)); do echo
> > CPU$i;
> > sudo rdmsr -p $i 0x771;  sudo rdmsr -p $i 0x774; done
> > 
> > But I expect them to match the acpi_cppc/highest_perf, which is
> > showing
> > 3.9GHz.
> > 
> > What command you to report " M-Test  . Id"?
> > 
> > I think some busy 100% workload running on a single CPU.
> > 
> > 
> > 
> > Analsis:
> > 
> > Intel® Core™ Ultra 9 Processor 275HX
> > 8+16 no HT
> > Performance-core Base Frequency : 2.7 GHz
> > Efficient-core Base Frequency : 2.1 GHz
> > 
> > 
> > P-cores: 0-7
> > Nominal freq: 2700
> > Nominal perf: 43
> > Scaling : ~1.6 (correct)
> > 
> > E-cores:
> > 8-
> > Nominal freq: 2100
> > Nominal perf: 29
> > Scaling : ~1.38 (close to 1.4 so there will be some rounding issue)
> > 
> > So,  this is not related to scaling factor as before like on ASROCK
> > 
> > 
> > The P-cores under report `base_frequency` and `cpuinfo_max_freq`.
> > 
> > - FAIL: With Turbo ON or off, the claimed `base_freqency` (B-Claim)
> > of
> > 2000000 does
> >   NOT match the Intel spec (B-Spec) of 2700000.
> > 
> > From CPPC:
> > 
> > /sys/devices/system/cpu/cpu0/acpi_cppc/guaranteed_perf:32
> > The base freq reported by cpufreq: 3200/1.6 = 2000
> > 
> > 
> > 
> > 
> > - FAIL: With Turbo off, the claimed `cpuinfo_max_freq` (M-Claim) of
> > 2000000 does
> >   NOT match the Intel spec (M-Spec) of 2700000.
> > 
> > We don’t depend on ACPI CPPC to achieve the measured value. When
> > you
> > are busy you are getting 2700, so the behavior is correct as the
> > per
> > the CPU spec.
> > 
> > 
> > 
> > - FAIL: With Turbo ON, the claimed `cpuinfo_max_freq` (M-Claim) of
> > 3900000 does
> >   NOT match the Intel spec (M-Spec) of 5400000.
> > 
> > ACPI CPPC also tells 3.9 GHz as max. But we don’t depend on it.
> 
> But we depend on HWP_CAP.highest_perf == CPPC.highest_perf (because
> the CPPC value comes from HWP).

That's why asked for dump of MSRs above. This should match CPPC value.
But seems setting HWP_REQ.max_perf=61 or 62 on P-core still results in
5.4 Ghz.

Thanks,
Srinivas



  reply	other threads:[~2025-12-05 19:50 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-05-18  3:33 [BUG] intel_pstate: CPU frequencies miscalculated/incorrectly detected on Arrow Lake hardware Aaron Rainbolt
2025-05-19 12:48 ` srinivas pandruvada
2025-11-06 19:46   ` Aaron Rainbolt
2025-07-22 16:31 ` Aaron Rainbolt
2025-07-22 17:24   ` srinivas pandruvada
2025-11-26 23:00     ` Aaron Rainbolt
2025-12-03 17:38       ` Aaron Rainbolt
2025-12-05 19:10         ` srinivas pandruvada
2025-12-05 19:28           ` Rafael J. Wysocki
2025-12-05 19:50             ` srinivas pandruvada [this message]
2025-12-05 20:04               ` Rafael J. Wysocki
2025-12-15  0:45           ` Aaron Rainbolt
2025-12-15 14:16             ` srinivas pandruvada
2025-12-15 15:25               ` Aaron Rainbolt
2025-12-17 13:33                 ` srinivas pandruvada
2025-12-23 17:41                   ` Aaron Rainbolt
2025-12-23 19:21                     ` srinivas pandruvada
2026-01-11  4:53 ` Russell Haley
2026-01-26 16:14   ` Aaron Rainbolt

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=c7bdb6cb02437d1ab41495d526c3d2ee3f1f7c60.camel@linux.intel.com \
    --to=srinivas.pandruvada@linux.intel.com \
    --cc=arainbolt@kfocus.org \
    --cc=kernel-team@lists.ubuntu.com \
    --cc=lenb@kernel.org \
    --cc=linux-acpi@vger.kernel.org \
    --cc=linux-pm@vger.kernel.org \
    --cc=mmikowski@kfocus.org \
    --cc=rafael@kernel.org \
    --cc=rjw@rjwysocki.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.